Show More
@@ -1,171 +1,170 | |||||
1 | # encoding: utf-8 |
|
1 | # encoding: utf-8 | |
2 |
|
2 | |||
3 | """Classes used in scattering and gathering sequences. |
|
3 | """Classes used in scattering and gathering sequences. | |
4 |
|
4 | |||
5 | Scattering consists of partitioning a sequence and sending the various |
|
5 | Scattering consists of partitioning a sequence and sending the various | |
6 | pieces to individual nodes in a cluster. |
|
6 | pieces to individual nodes in a cluster. | |
7 |
|
7 | |||
8 |
|
8 | |||
9 | Authors: |
|
9 | Authors: | |
10 |
|
10 | |||
11 | * Brian Granger |
|
11 | * Brian Granger | |
12 | * MinRK |
|
12 | * MinRK | |
13 |
|
13 | |||
14 | """ |
|
14 | """ | |
15 |
|
15 | |||
16 | #------------------------------------------------------------------------------- |
|
16 | #------------------------------------------------------------------------------- | |
17 | # Copyright (C) 2008-2011 The IPython Development Team |
|
17 | # Copyright (C) 2008-2011 The IPython Development Team | |
18 | # |
|
18 | # | |
19 | # Distributed under the terms of the BSD License. The full license is in |
|
19 | # Distributed under the terms of the BSD License. The full license is in | |
20 | # the file COPYING, distributed as part of this software. |
|
20 | # the file COPYING, distributed as part of this software. | |
21 | #------------------------------------------------------------------------------- |
|
21 | #------------------------------------------------------------------------------- | |
22 |
|
22 | |||
23 | #------------------------------------------------------------------------------- |
|
23 | #------------------------------------------------------------------------------- | |
24 | # Imports |
|
24 | # Imports | |
25 | #------------------------------------------------------------------------------- |
|
25 | #------------------------------------------------------------------------------- | |
26 |
|
26 | |||
27 | from __future__ import division |
|
27 | from __future__ import division | |
28 |
|
28 | |||
29 | import types |
|
29 | import types | |
30 | from itertools import islice |
|
30 | from itertools import islice | |
31 |
|
31 | |||
32 | from IPython.utils.data import flatten as utils_flatten |
|
32 | from IPython.utils.data import flatten as utils_flatten | |
33 |
|
33 | |||
34 | #------------------------------------------------------------------------------- |
|
34 | #------------------------------------------------------------------------------- | |
35 | # Figure out which array packages are present and their array types |
|
35 | # Figure out which array packages are present and their array types | |
36 | #------------------------------------------------------------------------------- |
|
36 | #------------------------------------------------------------------------------- | |
37 |
|
37 | |||
38 | arrayModules = [] |
|
38 | arrayModules = [] | |
39 | try: |
|
39 | try: | |
40 | import Numeric |
|
40 | import Numeric | |
41 | except ImportError: |
|
41 | except ImportError: | |
42 | pass |
|
42 | pass | |
43 | else: |
|
43 | else: | |
44 | arrayModules.append({'module':Numeric, 'type':Numeric.arraytype}) |
|
44 | arrayModules.append({'module':Numeric, 'type':Numeric.arraytype}) | |
45 | try: |
|
45 | try: | |
46 | import numpy |
|
46 | import numpy | |
47 | except ImportError: |
|
47 | except ImportError: | |
48 | pass |
|
48 | pass | |
49 | else: |
|
49 | else: | |
50 | arrayModules.append({'module':numpy, 'type':numpy.ndarray}) |
|
50 | arrayModules.append({'module':numpy, 'type':numpy.ndarray}) | |
51 | try: |
|
51 | try: | |
52 | import numarray |
|
52 | import numarray | |
53 | except ImportError: |
|
53 | except ImportError: | |
54 | pass |
|
54 | pass | |
55 | else: |
|
55 | else: | |
56 | arrayModules.append({'module':numarray, |
|
56 | arrayModules.append({'module':numarray, | |
57 | 'type':numarray.numarraycore.NumArray}) |
|
57 | 'type':numarray.numarraycore.NumArray}) | |
58 |
|
58 | |||
59 | class Map: |
|
59 | class Map(object): | |
60 | """A class for partitioning a sequence using a map.""" |
|
60 | """A class for partitioning a sequence using a map.""" | |
61 |
|
61 | |||
62 | def getPartition(self, seq, p, q): |
|
62 | def getPartition(self, seq, p, q): | |
63 | """Returns the pth partition of q partitions of seq.""" |
|
63 | """Returns the pth partition of q partitions of seq.""" | |
64 |
|
64 | |||
65 | # Test for error conditions here |
|
65 | # Test for error conditions here | |
66 | if p<0 or p>=q: |
|
66 | if p<0 or p>=q: | |
67 | print "No partition exists." |
|
67 | print "No partition exists." | |
68 | return |
|
68 | return | |
69 |
|
|
69 | ||
70 |
|
|
70 | N = len(seq) | |
71 | basesize = len(seq)//q |
|
71 | remainder = N % q | |
72 |
|
|
72 | basesize = N // q | |
73 |
|
|
73 | ||
74 | for n in range(q): |
|
74 | if p < remainder: | |
75 | if n < remainder: |
|
75 | low = p * (basesize + 1) | |
76 |
|
|
76 | high = low + basesize + 1 | |
77 | hi.append(lo[-1] + basesize + 1) |
|
77 | else: | |
78 | else: |
|
78 | low = p * basesize + remainder | |
79 | lo.append(n*basesize + remainder) |
|
79 | high = low + basesize | |
80 | hi.append(lo[-1] + basesize) |
|
|||
81 |
|
80 | |||
82 | try: |
|
81 | try: | |
83 |
result = seq[lo |
|
82 | result = seq[low:high] | |
84 | except TypeError: |
|
83 | except TypeError: | |
85 | # some objects (iterators) can't be sliced, |
|
84 | # some objects (iterators) can't be sliced, | |
86 | # use islice: |
|
85 | # use islice: | |
87 |
result = list(islice(seq, lo |
|
86 | result = list(islice(seq, low, high)) | |
88 |
|
87 | |||
89 | return result |
|
88 | return result | |
90 |
|
89 | |||
91 | def joinPartitions(self, listOfPartitions): |
|
90 | def joinPartitions(self, listOfPartitions): | |
92 | return self.concatenate(listOfPartitions) |
|
91 | return self.concatenate(listOfPartitions) | |
93 |
|
92 | |||
94 | def concatenate(self, listOfPartitions): |
|
93 | def concatenate(self, listOfPartitions): | |
95 | testObject = listOfPartitions[0] |
|
94 | testObject = listOfPartitions[0] | |
96 | # First see if we have a known array type |
|
95 | # First see if we have a known array type | |
97 | for m in arrayModules: |
|
96 | for m in arrayModules: | |
98 | #print m |
|
97 | #print m | |
99 | if isinstance(testObject, m['type']): |
|
98 | if isinstance(testObject, m['type']): | |
100 | return m['module'].concatenate(listOfPartitions) |
|
99 | return m['module'].concatenate(listOfPartitions) | |
101 | # Next try for Python sequence types |
|
100 | # Next try for Python sequence types | |
102 | if isinstance(testObject, (types.ListType, types.TupleType)): |
|
101 | if isinstance(testObject, (types.ListType, types.TupleType)): | |
103 | return utils_flatten(listOfPartitions) |
|
102 | return utils_flatten(listOfPartitions) | |
104 | # If we have scalars, just return listOfPartitions |
|
103 | # If we have scalars, just return listOfPartitions | |
105 | return listOfPartitions |
|
104 | return listOfPartitions | |
106 |
|
105 | |||
107 | class RoundRobinMap(Map): |
|
106 | class RoundRobinMap(Map): | |
108 | """Partitions a sequence in a roun robin fashion. |
|
107 | """Partitions a sequence in a roun robin fashion. | |
109 |
|
108 | |||
110 | This currently does not work! |
|
109 | This currently does not work! | |
111 | """ |
|
110 | """ | |
112 |
|
111 | |||
113 | def getPartition(self, seq, p, q): |
|
112 | def getPartition(self, seq, p, q): | |
114 | # if not isinstance(seq,(list,tuple)): |
|
113 | # if not isinstance(seq,(list,tuple)): | |
115 | # raise NotImplementedError("cannot RR partition type %s"%type(seq)) |
|
114 | # raise NotImplementedError("cannot RR partition type %s"%type(seq)) | |
116 | return seq[p:len(seq):q] |
|
115 | return seq[p:len(seq):q] | |
117 | #result = [] |
|
116 | #result = [] | |
118 | #for i in range(p,len(seq),q): |
|
117 | #for i in range(p,len(seq),q): | |
119 | # result.append(seq[i]) |
|
118 | # result.append(seq[i]) | |
120 | #return result |
|
119 | #return result | |
121 |
|
120 | |||
122 | def joinPartitions(self, listOfPartitions): |
|
121 | def joinPartitions(self, listOfPartitions): | |
123 | testObject = listOfPartitions[0] |
|
122 | testObject = listOfPartitions[0] | |
124 | # First see if we have a known array type |
|
123 | # First see if we have a known array type | |
125 | for m in arrayModules: |
|
124 | for m in arrayModules: | |
126 | #print m |
|
125 | #print m | |
127 | if isinstance(testObject, m['type']): |
|
126 | if isinstance(testObject, m['type']): | |
128 | return self.flatten_array(m['type'], listOfPartitions) |
|
127 | return self.flatten_array(m['type'], listOfPartitions) | |
129 | if isinstance(testObject, (types.ListType, types.TupleType)): |
|
128 | if isinstance(testObject, (types.ListType, types.TupleType)): | |
130 | return self.flatten_list(listOfPartitions) |
|
129 | return self.flatten_list(listOfPartitions) | |
131 | return listOfPartitions |
|
130 | return listOfPartitions | |
132 |
|
131 | |||
133 | def flatten_array(self, klass, listOfPartitions): |
|
132 | def flatten_array(self, klass, listOfPartitions): | |
134 | test = listOfPartitions[0] |
|
133 | test = listOfPartitions[0] | |
135 | shape = list(test.shape) |
|
134 | shape = list(test.shape) | |
136 | shape[0] = sum([ p.shape[0] for p in listOfPartitions]) |
|
135 | shape[0] = sum([ p.shape[0] for p in listOfPartitions]) | |
137 | A = klass(shape) |
|
136 | A = klass(shape) | |
138 | N = shape[0] |
|
137 | N = shape[0] | |
139 | q = len(listOfPartitions) |
|
138 | q = len(listOfPartitions) | |
140 | for p,part in enumerate(listOfPartitions): |
|
139 | for p,part in enumerate(listOfPartitions): | |
141 | A[p:N:q] = part |
|
140 | A[p:N:q] = part | |
142 | return A |
|
141 | return A | |
143 |
|
142 | |||
144 | def flatten_list(self, listOfPartitions): |
|
143 | def flatten_list(self, listOfPartitions): | |
145 | flat = [] |
|
144 | flat = [] | |
146 | for i in range(len(listOfPartitions[0])): |
|
145 | for i in range(len(listOfPartitions[0])): | |
147 | flat.extend([ part[i] for part in listOfPartitions if len(part) > i ]) |
|
146 | flat.extend([ part[i] for part in listOfPartitions if len(part) > i ]) | |
148 | return flat |
|
147 | return flat | |
149 | #lengths = [len(x) for x in listOfPartitions] |
|
148 | #lengths = [len(x) for x in listOfPartitions] | |
150 | #maxPartitionLength = len(listOfPartitions[0]) |
|
149 | #maxPartitionLength = len(listOfPartitions[0]) | |
151 | #numberOfPartitions = len(listOfPartitions) |
|
150 | #numberOfPartitions = len(listOfPartitions) | |
152 | #concat = self.concatenate(listOfPartitions) |
|
151 | #concat = self.concatenate(listOfPartitions) | |
153 | #totalLength = len(concat) |
|
152 | #totalLength = len(concat) | |
154 | #result = [] |
|
153 | #result = [] | |
155 | #for i in range(maxPartitionLength): |
|
154 | #for i in range(maxPartitionLength): | |
156 | # result.append(concat[i:totalLength:maxPartitionLength]) |
|
155 | # result.append(concat[i:totalLength:maxPartitionLength]) | |
157 | # return self.concatenate(listOfPartitions) |
|
156 | # return self.concatenate(listOfPartitions) | |
158 |
|
157 | |||
159 | def mappable(obj): |
|
158 | def mappable(obj): | |
160 | """return whether an object is mappable or not.""" |
|
159 | """return whether an object is mappable or not.""" | |
161 | if isinstance(obj, (tuple,list)): |
|
160 | if isinstance(obj, (tuple,list)): | |
162 | return True |
|
161 | return True | |
163 | for m in arrayModules: |
|
162 | for m in arrayModules: | |
164 | if isinstance(obj,m['type']): |
|
163 | if isinstance(obj,m['type']): | |
165 | return True |
|
164 | return True | |
166 | return False |
|
165 | return False | |
167 |
|
166 | |||
168 | dists = {'b':Map,'r':RoundRobinMap} |
|
167 | dists = {'b':Map,'r':RoundRobinMap} | |
169 |
|
168 | |||
170 |
|
169 | |||
171 |
|
170 |
General Comments 0
You need to be logged in to leave comments.
Login now